Tamm Review: On the nature of the nitrogen limitation to plant growth in Fennoscandian boreal forests

(1)

Tamm Review: On the nature of the nitrogen limitation to plant growth in Fennoscandian boreal forests

Peter Högberg

^a

, Torgny Näsholm

^a

, Oskar Franklin

^a,b

, Mona N. Högberg

^a,^⇑

aDepartment of Forest Ecology and Management, Swedish University of Agricultural Sciences, SE-901 83 Umeå, Sweden

bInternational Institute for Applied Systems Analysis (IIASA), A-2361, Laxenburg, Austria

a r t i c l e i n f o

Article history:

Received 5 January 2017

Received in revised form 18 April 2017 Accepted 22 April 2017

Available online xxxx Keywords:

Boreal forests Forest management Forest production Soil microbial community Mycorrhiza

Nitrogen cycling

a b s t r a c t

The supply of nitrogen commonly limits plant production in boreal forests and also affects species composition and ecosystem functions other than plant growth. These interrelations vary across the landscapes, with the highest N availability, plant growth and plant species richness in ground-water discharge areas (GDAs), typically in toe-slope positions, which receive solutes leaching from the much larger groundwater recharge areas (GRAs) uphill. Plant N sources include not only inorganic N, but, as heightened more recently, also organic N species. In general, also the ratio inorganic N over organic N sources increase down hillslopes. Here, we review recent evidence about the nature of the N limitation and its variations in Fennoscandian boreal forests and discuss its implications for forest ecology and management.

The rate of litter decomposition has traditionally been seen as the determinant of the rate of N supply.

However, while N-rich litter decomposes faster than N-poor litter initially, N-rich litter then decomposes more slowly, which means that the relation between N % of litter and its decomposability is complex.

Moreover, in the lower part of the mor-layer, where the most superficial mycorrhizal roots first appear, and N availability matters for plants, the ratio of microbial N over total soil N is remarkably constant over the wide range in litter and soil C/N ratios of between 15 and 40 for N-rich and N-poor sites, respectively.

Nitrogen-rich and -poor sites thus differ in the sizes of the total N pool and the microbial N pool, but not in the ratio between them. A more important difference is that the soil microbial N pool turns over faster in N-rich systems because the microbes are more limited by C, while microbes in N-poor systems are a stronger sink for available N.

Furthermore, litter decomposition in the most superficial soil horizon (as studied by the so-called litter-bag method) is associated with a dominance of saprotrophic fungi, and absence of mycorrhizal fungi. The focal zone in the context of plant N supply in N-limited forests is further down the soil profile, where ectomycorrhizal (ECM) roots become abundant. Molecular evidence and stable isotope data indi- cate that in the typical N-poor boreal forests, nitrogen is retained in saprotrophic fungi, likely until they run out of energy (available C-compounds). Then, as heightened by recent research, ECM fungi, which are supplied by photosynthate from the trees, become the superior competitors for N.

In N-poor boreal soils strong N retention by microorganisms keeps levels of available N very low. This is exacerbated by an increase in tree C allocation to mycorrhizal fungi (TCAM) relative to net primary production (NPP) with decreasing soil N supply, which causes ECM fungi to retain much of the available soil N for their own growth and transfer little to their tree hosts. The transfer of N through the ECM fungi, and not the rate of litter decomposition, is likely limiting the rate of tree N supply under such conditions. All but a few stress-tolerant less N-demanding plant species, like the ECM trees themselves and ericaceous dwarf shrubs, are excluded.

With increasing N supply, a weakening of ECM symbiosis caused by the relative decline in TCAM con- tributes to shifts in soil microbial community composition from fungal dominance to bacterial dominance. Thus, bacteria, which are less C-demanding, but more likely to release N than fungi, take over.

This, and the relatively high pH in GDA, allow autotrophic nitrifying bacteria to compete successfully for the NH4+released by C-limited organisms and causes the N cycle to open up with leaching of nitrate (NO3) and gaseous N losses through denitrification. These N-rich conditions allow species-rich communities of N-demanding plant species. Meanwhile, ECM fungi have a smaller biomass, are supplied with N in excess of their demand and will export more N to their host trees. Hence, the gradient from low to high

http://dx.doi.org/10.1016/j.foreco.2017.04.045

0378-1127/Ó2017 The Authors. Published by Elsevier B.V.

This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

⇑ Corresponding author.

Contents lists available atScienceDirect

Forest Ecology and Management

j o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / f o r e c o

Please cite this article in press as: Högberg, P., et al. Tamm Review: On the nature of the nitrogen limitation to plant growth in Fennoscandian boreal for-

(2)

N supply is characterized by profound variations in plant and soil microbial physiologies, especially their relations to the C-to-N supply ratio. We propose how interactions among functional groups can be under- stood and modelled (the plant-microbe carbon-nitrogen model).

With regard to forest management these perspectives explain why the creation of larger tree-free gaps favors the regeneration of tree seedlings under N-limited conditions through reduced belowground competition for N, and why such gaps are less important under high N supply (but when light might be limiting). We also discuss perspectives on the relations between N supply, biodiversity, and eutrophication of boreal forests from N deposition or forest fertilization.

Ó2017 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Contents

1. Introduction . . . 00

2. Background: Previous observations of and hypotheses about variations in N cycling in Fennoscandian boreal and other similar forests . . . 00

2.1. Early observations . . . 00

2.2. N cycling – classical perspectives. . . 00

2.3. Towards a more diverse perspective on N availability . . . 00

3. Nitrogen supply to and within boreal forests. . . 00

3.1. Which are the inputs and how large are they? . . . 00

3.2. How important are plant litter characteristics? . . . 00

3.3. How is N released in available forms and which forms of N are used by microbes and plants?. . . 00

3.4. Methodological limitations and developments . . . 00

3.5. Where in the soil profile is N supplied to trees and other plants? . . . 00

4. Variations in N supply in boreal landscapes. . . 00

4.1. Causes of variations in plant N supply in relation to hill-slope hydrology: abiotic perspectives. . . 00

4.2. How important are soil pH and N supply for microbial release of N, and which is the role of mycorrhizal fungi (biotic perspectives)? . 00 4.3. May other elements than N limit plant growth? . . . 00

5. Interactions among trees, microbes and soil: The Plant-Microbe Carbon-Nitrogen model . . . 00

5.1. On the C for N exchange rates in mycorrhizal symbiosis . . . 00

5.2. Plant-microbe interactions – implications for N cycling. . . 00

6. Implications for forest ecology and management . . . 00

6.1. Should all forests be regenerated by the same method (clear-felling vs. continuous-cover forestry)? . . . 00

6.2. A few remarks on leakage of N from boreal forests . . . 00

6.3. Can a self-sustaining greater N cycle be induced by boost additions of N? . . . 00

6.4. How will global changes affect the Fennoscandian boreal forests? Will we see N enrichment through N deposition or progressive N limitation (PNL) driven by increasing [CO2]? . . . 00

6.5. On the relations among N supply, plant biodiversity and plant growth . . . 00

7. Conclusions and suggestions for future research . . . 00

Acknowledgements . . . 00

References . . . 00

1. Introduction

Boreal forests cover vast areas in the northern circumpolar region (Chapin et al., 2011), which is characterized by long cold winters and short summers. Another important feature is that plant growth is commonly constrained by a low supply of N, i.e.

plant growth is enhanced by additions of N (e.g.,Tamm, 1991).

Tree growth varies with hill-slope position (e.g., Hägglund and Lundmark, 1977), likely because of effects of hill-slope hydrology on the availability of N. Variations in N supply also affects the species composition and other functions of boreal forests.

People dependent on products from the forests and the possibility to clear them for the purpose of agriculture (permanently or temporarily as swidden agriculture) probably recognized already centuries ago the existence of relations between certain plant species and soil fertility. A first more systematic description linking the common forest types (based on descriptions of the field-layer plants) to forest productivity in Finland was given by Cajander (1909, 1926) and Cajander and Ilvessalo (1922). According to these schemes, and followers from Sweden (Arnborg, 1990) and Norway

(Kielland-Lund, 1982), ericaceous dwarf shrub communities cover the poorer soils, short herbs are common at intermediate soil fer- tilities and tall herbs dominate richer soils. The exact causes of these differences in species composition and productivity were not clarified and there is still no commonly accepted explanation of their nature.

This review serves the purpose of describing these patterns and their likely causes in the perspective of recent advances in our understanding of the interactions among soils, soil microorganisms, and plants. We will focus on boreal forests on mineral soils (i.e. make little reference to forests on peat) in Fennoscandia. How- ever, the patterns and processes discussed may be important also in other N-limited forests, e.g., other boreal and temperate forests.

For example, forest growth on drained boreal peatlands can like- wise be predicted based on the composition of the original field- layer plant community (Hånell, 1988). Moreover, an understanding of these patterns and processes and their underlying causes is essential not only for ecologists interested in the structure and function of these ecosystems, but also pertinent to aspects on forest management (e.g.,Hynynen et al., 2005; Mäkinen et al., 2006).

(3)

2. Background: Previous observations of and hypotheses about variations in N cycling in Fennoscandian boreal and other similar forests

2.1. Early observations

In the early 20th century, it was well established that the supply of N often limited the production of agricultural crops. The first straightforward N addition experiment in Swedish forests made by Hesselman in the early 1920s indicated that this limitation may also prevail in forests (Hesselman, 1926). This has subsequently received massive support by N fertilizer trials across Fennoscandia (e.g.,Nilsen, 2001; Nohrstedt, 2001; Saarsalmi and Mälkönen, 2001). Moreover, already in 1917 Hesselman reported the occurrence of nitrate in leaves of herbs in lush tall herb vegetation found in ‘‘lunddälder” (Hesselman, 1917). ‘‘Lund” refers to vegetation dominated by deciduous trees and herbs (and sometimes broad-leaved grasses) and ‘‘dälder” is plural for small valleys, i.e.

sites likely to receive groundwater discharge, at least temporarily.

Hesselman assumed that the NO3in the leaves was not produced by the plants (which indeed is very rare, and is only known to occur in a few N2-fixing legume species, seeHipkin et al., 2004), but must have been taken up from the soil. He also noted that NO3 occurred in leaves of some plant species in clear-fellings. Corroborating observations were later made using anin vivomethod for measur- ing leaf nitrate reductase activity (NRA), which revealed increasing NRA along a gradient from ericaceous dwarf shrub type, through a short herb type to a tall herb type (Högberg et al., 1990) and ele- vated NRA in clear-fellings (Högbom et al., 2002).

The early observations of spatial variations in forest plant growth stimulated soil studies aiming at unravelling the underlying causes. Ambitious surveys and studies were conducted in Fin- land by Viro (1951, 1955), who found a correlation between exchangeable Ca²⁺and forest productivity. After re-examining this material,Dahl et al. (1961)concluded that the correlation was not unequivocally supporting the notion that Ca was limiting forest growth. They made a detailed regional survey in Hedmark County, Norway, of relations between soil conditions and plant community composition (Dahl et al., 1967). The survey showed very strong correlations among forest vegetation types (classification based on field-layer species), forest growth and soil variables like base saturation (which is dominated by the contribution of exchangeable Ca²⁺), but also N concentration. The latter was, of course, particularly interesting given the then well-established N limitation to forest growth.

Subsequently,Lahti and Väisänen (1987)made use of vegetation and soils data from previous surveys in Finnish southern boreal forests. They found striking correlations among forest vegetation types and soil pH, exchangeable Ca²⁺, and % N, i.e. variables known to correlate with forest growth. Remarkable variations in such relations between soils and vegetation can be found in short distances, as demonstrated byGiesler et al. (1998)in a study of a 90- m-long transect from a groundwater recharge area (GRA) to a discharge area (GDA), along which forest production varied by a factor three. Furthermore, survey data from Finland (Lahti and Väisänen, 1987) and Norway (Dahl et al., 1967), and the sharp local boreal Swedish gradient (Giesler et al., 1998) fits almost the same regression between % N and exchangeable Ca²⁺(Fig. 1a). With comparable forest types in the three countries, they could be ordered from poor to rich forest types (i.e. from dwarf shrub types to tall herb types) along this positive relation (Giesler et al., 1998, seeFig. 1b). Simi- larly, there is a very strong negative correlation between the soil C/N ratio (which, of course, is very strongly negatively correlated with % N in the organic mor-layer) and base saturation, BS, as shown by data from the Swedish Forest Soil Survey (Fig. 2).

2.2. N cycling – classical perspectives

So, how do the N cycles and N release patterns differ among forest types? A classical paradigm assumes there is a positive relationship between litter decomposability (loss of litter mass) and plant N availability and focuses on effects of litter quality (Bosatta and Ågren, 1991; Ågren and Bosatta, 1996), i.e. primarily the C/N ratio, and on abiotic constraints (e.g., temperature, moisture), on the rate of litter decomposition (e.g.,Swift et al., 1979;

Berg and McClaugherty, 2008).

According to studies of decomposition of litter enclosed in so called litter bags, a widely used method, the rate of decomposition is initially (for the first few years) higher in N-rich litter, but then becomes slower than in N-poor litter (Berg and McClaugherty, 2008). Hence, the relation between the release of N in plant available forms and the C/N ratio of the litter is not straightforward. In addition to the abiotic constraints, moisture and temperature, the amount of plant available N is also dependent on the amount of litter deposited, and on the strength of competing biotic and abiotic sinks for available N. A crucial question to be discussed further below (see Section3.2.) is if a change of plant species composition, and hence litter characteristics, can cause a substantial change in the nature of the N cycle.

Fig. 1.The relation between N concentration and base saturation in mor-layers in (a) Finland (Lahti and Väisänen, 1987), Hedmark county, Norway (Dahl et al., 1967), and along the Betsele transect in Sweden (Giesler et al., 1998). (b) The relationship between % N and % base saturation (BS) in the mor layer in dwarf shrub (DS), short herb (SH), and tall herb (TH) forest types in Finland, Sweden, and Norway. Data from Finland (Lahti and Väisänen, 1987) represent 921 forest stands. Base saturation data fromGiesler et al. (1998)andDahl et al. (1967)is determined in unbuffered 1.0 mol/L NH4NO3and NH4C2H3O2, respectively, whereas the Finnish material represents Ca saturation (extraction method unknown). The difference should be minor because of the dominant role of Ca²⁺among base cations.

(4)

One line of reasoning has focused on the possibility that much of the initial plant litter N deposited is not released by decomposers, but perhaps ultimately forms the more stable N compounds deeper down in the soil. To this, depolymerization- recondensation pathways or various other chemical immobilization or selective preservation pathways could contribute in making soil organic N less available to decomposers (Stevenson, 1994;

Knicker, 2011). There is an array of organic N forms that could be formed through reactions between inorganic or organic N sources and lignin compounds or other phenolic compounds. The traditional view on causes to the stability of soil organic matter C and N focused on the chemical properties of molecules per se, and stable organic matter was thus said to be formed through the for- mation of humic substances or through selective preservation of resistant components (Stevenson, 1994).

Over time, the view has changed to a continuum of many different organic compounds at various stages of decomposition (Bosatta and Ågren, 1991) and the emerging understanding is that most old soil organic matter can be further decomposed (Knicker, 2011; Schmidt et al., 2011; Lehmann and Kleber, 2015). It has also been proposed that this material, while of plant origin initially, consists of material proximally derived from microorganisms (Gleixner, 2013). However, with time smaller and less energy- rich compounds remain (Bosatta and Ågren, 1991) and the supply of available C to the decomposer microorganisms diminishes.

Recent research shows that much of the older soil organic N consists of peptide-like compounds, which (unlike the original litter N) have the potential to be selectively preserved during further decomposition and to dominate the soil organic N pool (Knicker and Kögel-Knabner, 1998; Knicker, 2011).

Studying plant available N is complex, because N occurs in many forms in the soil and most of the N occurs in compounds turning over very slowly, whereas the active pools are very small and difficult to measure (Binkley and Hart, 1989). For example, in a boreal forest mor-layer, extractable inorganic N (NH4+ and NO3) and extractable (using KCl as extractant) amino acids and microbial N contributed 1‰, 3‰and 10%, respectively, of the total

N (Näsholm et al., 1998). These small pools turn over quickly, days for inorganic N and weeks for microbial N (Högberg et al., 2006), as a result of rapid biological mobilization and immobilization processes, lysis of microorganisms and grazing on them by soil animals.

For long, the major methods to asses plant available N have been biological (so called bio-assays) and based on incubation of samples in the field or in the laboratory, under conditions that should promote N mineralization from organic sources (e.g., Binkley and Hart, 1989). Chemical extraction (usually with a solution of KCl or K2SO4as extractant) at intervals after soil sampling is followed by analysis of inorganic N in the filtered extracts. These widely used methods are based on the assumptions that the same biological processes that cause release of plant available N under undisturbed conditions in the field are also responsible for the production of inorganic N in the laboratory procedures (Bundy and Meisinger, 1994and references therein) and that plant available N is restricted to inorganic N forms. In agricultural soils, for which these biological methods to estimate net N mineralization was first developed, there is usually a rapid net release of inorganic N (Jansson, 1958).

2.3. Towards a more diverse perspective on N availability

In contrast to the situation in arable soils, there is often no net N mineralization or net nitrification during the first months of incubation in most boreal and other high-latitude soils (e.g.,Priha and Smolander, 1999; Persson et al., 2000; Stark et al., 2003). However, trees and other plants apparently do take up N also under these cir- cumstances. Discrepancies between net N mineralization and plant N uptake stimulated research on uptake of organic N sources (Chapin et al., 1993; Kielland, 1994) and measurements of gross (actual) N mineralization (see below) and gross nitrification rates (e.g., Davidson et al., 1991, 1992; Binkley et al., 1992a; Hart et al., 1994; Stark and Hart, 1997; Fisk and Fahey, 2001; Merilä et al., 2002; Stark et al., 2003; Högberg et al., 2006). Such studies indeed showed measurable mineralization rates also in the shorter term (see below); supporting the idea that inorganic N could be a source, albeit not the single source of plant N in boreal forests.

Which could the N sources be if the supply of inorganic N from decomposition of organic matter seems too low? Trees ofAlnus forms symbiosis with N2-fixing actinobacteria, from which they gain N (Johnsrud, 1978; Huss-Danell, 1997), but other boreal tree species lack this ability. This means that the ‘‘missing” sources of N for the other boreal tree species could be organic N forms, or perhaps inorganic N is still important, but release and uptake are so tightly coupled that free inorganic N cannot be observed. The latter is possible if the microbial community remains a strong N sink even under long laboratory soil incubations (Hart et al., 1994). This condition changes dramatically when the microorganisms exploit- ing a specific ageing substrate over time become C-limited, with increased gross N and even net N mineralization as a result (Hart et al., 1994) in line with our current understanding of microbial physiology (Franklin et al., 2011; Robertson and Groffman, 2015).

In C-limited soil many soil microorganisms are found close to roots as a result of exudation of highly degradable C substrates (Nazir et al., 2010, and references therein); the high C supply can support a microbial N sink strength and, hence, contribute to a low N availability for plants. Gadgil and Gadgil (1971, 1975) showed in a study on New Zealand that decomposition could become greater on trenched plots excluding the influence of active ECM roots. They ascribed the higher rate of decomposition to a higher availability of N. Subsequent studies of the so called ‘‘Gadgil effect” have, however, been inconsistent (Fernandez and Kennedy, 2016).

Fig. 2.The relation between mor-layer C/N ratio and base saturation for forests in Norrland, Sweden. Data are from the Forest Soil Survey 1993–2002 and are means for different vegetation types. The black circles are contributed by 2587 samples, which represent 79% of the total number of samples (n = 3264 from an area of 12.45 million ha of forests). An additional 18% were contributed by forest types dominated by forest types characterized by grasses and sedges (white unfilled circles), which indicates that they may have been used for crops or as pastures. We have excluded rare forest types, with less than 10 samples each. The DS, SH and TH forest types (filled circles) contribute 79%, 12% and 9%, respectively, to the data (means). The regression line is based on all data points in the graph (R²adj= 0.85, n = 17, p < 0.001).

(5)

Alternatively, as proposed in other studies, the higher C supply and microbial biomass in the rhizosphere can stimulate N mineralization of available organic N sources and N immobilization processes (e.g., Norton and Firestone, 1991, 1996; Norton et al., 1990), and can accelerate the decomposition of soil organic matter for additional N mineralization, so called real priming (Kuzyakov and Xu, 2013). Similarly,Clarholm (1985)stressed the importance of a microbial loop, in which bacterial activity was stimulated by root exudation of C, which in turn fueled bacterial breakdown of complex N sources and plant assimilation of the N released after amoebae grazing on the bacteria. If this is the case, nitrogen mineralization should be lower in soil samples without active plant roots.

The rapid turnover of small pools of N in the soil can be studied by the¹⁵N pool dilution method (Kirkham and Bartholomew, 1954;

Jansson, 1958; Davidson et al., 1991, 1992). This method uses measurements of the change in pool size and labelling of the N pool studied. In studies of gross N mineralization, a trace of NH4+highly enriched in¹⁵N is injected into the soil, whereas the NO3 pool is labelled if gross nitrification is studied. In both cases the assump- tion is that the endogenous N source, from which the N in the studied inorganic pool is produced, is unlabeled, i.e. has an atom %¹⁵N at natural abundance. The method was introduced in agricultural settings in the 1950s (Kirkham and Bartholomew, 1954; Jansson, 1958), but was first used in forests in the late 1980s (e.g., Davidson et al., 1991, 1992). There, a common problem is the much smaller sizes of the inorganic pools of N. Interestingly, however, the method revealed rapid turnover of the pools of NH4+and sometimes also NO3 (Hart et al., 1994; Stark and Hart, 1997; Högberg et al., 2006) in forest soils, where these pools were small and where soil incubations for the monitoring of net release of these ions found very low rates of release if any release at all (Davidson et al., 1992; Högberg et al., 2006).

Most ecologists interested in N uptake by plants assumed for long that the plants take up N only in inorganic forms, despite the fact that plant physiologist working in parallel in laboratories observed plant uptake and growth on amino acid N (Hutchinson and Miller, 1912; Brigham, 1917; Virtanen and Linkola, 1946).

Moreover, uptake of¹⁵N in glutamic acid by a mycorrhizal fungus (grown under otherwise sterile conditions) transferring the¹⁵N to its plant hostPinus sylvestriswas reported in a laboratory study already in the early 1950s (Melin and Nilsson, 1953). However, it remained until the 1990s before plant growth on amino acids was revisited (Chapin et al., 1993) and uptake of an intact amino acid was demonstrated directly in the field in a boreal forest (Näsholm et al., 1998).

Before that, the idea of organic N as an important N source for boreal forest plants, in particular, was explored in laboratory studies byRead (1983, 1986, 1991) and his co-workers (Abuzinadah et al., 1986; Abuzinadah and Read, 1986). Read made reference to the observations that ectomycorrhizal (ECM) trees and ericaceous dwarf shrubs with ericoid mycorrhizas (ERM) were the dominant plants of boreal forests in America, Europe and Asia, and that this dominance was associated with N-poor conditions and forma- tion of acid organic mor-layers on top of the mineral soil. Further- more, based on the high capacity of many ECM and ERM fungi to take up organic N sources, Read and co-workers proposed that the dominances of the host plant species was linked to this capacity (Abuzinadah and Read, 1986; Abuzinadah et al., 1986). How- ever, a screening including AM, ERM, ECM and non-mycorrizal boreal forest plants did not show any difference in uptake rates of organic N in the form of amino acids (Persson and Näsholm, 2001). Moreover, detailed studies of the model plantArabidopsis thalianaindicates that even this ruderal, non-mycorrhizal plant has a well-developed capacity for organic N nutrition (Forsum et al., 2008). A widespread capacity to take up organic N is also

expected from an organism nutrition perspective because of the lower C cost and associated higher N-use efficiency of growth compared to uptake of inorganic N (Franklin et al., 2017).

In Read’s model he also included that more southerly forests and steppe vegetation is associated with a more complete decomposition of organic matter, a greater supply of N, but less of P, and a greater prominence (nemoral forests) or total dominance of plants (steppe) with arbuscular mycorrhiza (AM) (Read, 1991; Read and Perez-Moreno, 2003). Such variations from dominance of ECM and ERM associations to dominance of AM associations among field-layer plants can occur locally in the context of short (90-m- long) transects from recharge areas to discharge areas in boreal forests (Högberg et al., 1990; Giesler et al., 1998; Nilsson et al., 2005). The release of available N by mycorrhizal fungi from complex N sources represents another microbial loop, in which some of the N goes directly to the plant, some remains in the mycelium, but can become more freely available after grazing byCollembola andAcarion the mycorrhizal mycelium.

With this as background, we would like to turn to greater detail of what we now believe we know about the nature of the N limitation to plant growth that prevails in most boreal forests. Atten- tion will be drawn to what we consider critical new insights and emerging reformulations of old questions in the contexts of forest ecology and management in Fennoscandia.

3. Nitrogen supply to and within boreal forests 3.1. Which are the inputs and how large are they?

Minerals very rarely contain significant amounts of N, which can become available for plants as a result of weathering. Excep- tions are minerals with NH₄⁺fixed in crystal lattices, in which case studies in N. America have shown that weathering may even release enough NH4+to support high rates of nitrification and soil acidification (Dahlgren, 1994; Morford et al., 2016). Bedrocks of the Caledonian overthrust forming the Scandes Mts. contain traces of N (Dixon et al., 2012), but estimates of natural weathering are uncertain as well as of contributions of this source of N to the plants. As far as we know, there are no reports of release of N from the forested Fennoscandian bedrock shield, which is dominated by granites and gneisses. This leaves atmospheric N deposition and biological N2fixation as major known external sources of N to soil organisms and plants.

Sampling of the ice sheet of Greenland has revealed that there was some deposition of NO3 even before the impact of industrial- ization (Mayewski et al., 1986), when the deposition of this com- pound doubled. In Europe the increase was larger, but is now slowly declining (Lajhta and Jones, 2013). There has probably also always been a minor background contribution of N in dust from desert storms and aerosols of marine origin. As regards northern Fennoscandian boreal forests the total ambient N deposition ranges between 1 and 3 kg N ha ¹yr ¹ (Sponseller et al., 2016) and decreases from the SW to the NE.

Dinitrogen fixation is carried out by several genera of prokary- otic organisms functioning along a continuum from those living freely in soil to more intimate associations with other microorganisms and plants. In highly developed symbioses, the plant hosts supply the N2-fixing bacteria with the energy sources (C- compounds) needed to break the strong triple bonds of N2 and reduce it to NH3to support growth of the bacteria and the plant.

Highly developed symbioses are in boreal forests represented by actinobacteria,Frankia, in root nodule symbiosis with woody species likeAlnus incanaandMyrica gale, and by lichens forming tri- partite symbioses between fungi, green algae, and N2-fixing cyanobacteria; typical examples of which are the lichens Please cite this article in press as: Högberg, P., et al. Tamm Review: On the nature of the nitrogen limitation to plant growth in Fennoscandian boreal for-

(6)

Stereocaulon paschale, Peltighera aphtosaand Nephroma arcticum.

Grey alder (A. incana) occurs in a narrow zone of some tens of meters on recently exposed ground (due to iso-static rebound after the last glaciation) along the seashore of the Gulf of Bothnia, banks of rivers, in ditches along roads and occasionally as clones in boreal conifer forests. Across a primary chronosequence studied byBlaško et al. (2015) a build-up rate of soil N of 20 kg N ha ¹yr ¹ occurred in the narrow zone of alder forests, which may possibly fix more N2 (e.g., Johnsrud, 1978; Binkley et al., 1992b; Huss- Danell et al., 1992). Grey alder may also occur locally after clear- felling, and was used in swidden agriculture in Finland (Mikola et al., 1983). Leguminous plants are rare in boreal forests, but may occur after fire; we are unaware of estimates of their N2- fixation under such conditions.

Rates of N2fixation in lichen symbioses are lower than in acti- norhizal symbiosis; in a field study of a boreal pine forestHuss- Danell (1977) reported, based on acetylene reduction activity (ARA), a contribution by Stereocaulon paschale of < 2 kg N ha ¹- year ¹. She also found that ARA in the lichen was very sensitive to dry conditions (Huss-Danell, 1977). With regard to N₂-fixing bacteria more loosely associated with plants, most interest has recently been on cyanobacteria associated with mosses (e.g.,Basilier, 1979;

DeLuca et al., 2002). The contribution by the moss-associated N₂ fixation in boreal forest is estimated to 1–3 kg N ha ¹ year ¹ (DeLuca et al., 2002; Lindo et al., 2013; Stuiver et al., 2015). Hetero- trophic free living N2-fixing bacteria in soil are able to fix at a rate of 1 kg N ha ¹yr ¹(Nohrstedt, 1985).

Recent studies suggest endophytic bacteria may also contribute to tree N acquisition. Dinitrogen-fixing bacteria have been identified in tissues such as roots, needles and stems in some tree species (Moyes et al., 2016, and references therein), but the extent to which these endophytes contribute to the N nutrition of their host trees is unknown. In a¹³N labelling study it was shown that endophytes of Pinus needles fixed N at rates < 0.14 g N ha ¹day ¹ (Moyes et al., 2016), i.e. 28 g N ha ¹ yr ¹ if the active season is assumed to be 200 days, which suggest minor contributions to plant N supply. The genera ofRhizobiumandBradyrhizobiumhave been observed in association with ECM fungi in mycorrhizas as well as fungal sporocarps (Barbieri et al., 2005; Nguyen and Bruns, 2015), and were found in soils of natural N rich as well as N-fertilized boreal forests (Högberg et al., 2014a). Dinitrogen fixation (by bacteria) has also been detected in Suillus tomentosus tuberculate ECM, i.e. densely packed clusters of mycorrhizas (Paul et al., 2007). Again, it is not fully known to what extent these more recent findings of plant microbial associations enhance plant N supply.

The summed ecosystem N input and also output fluxes (Sponseller et al., 2016;Table 1) are throughout smaller than the internal cycling of N annually in boreal forests, with large losses in forest fires as exceptions. Gross N mineralization estimates which correct for microbially assimilated inorganic N, commonly show rates between 10 and 100 mg N m ²day ¹, although rates above 400 mg N m ² day ¹have been observed at exceptionally productive sites (Högberg et al., 2006; Blaško et al., 2015;Table 2).

Table 2

Chemical and biological characteristics of mor-layer soil (F + H horizons) along the 90-m-long N supply gradient from a dwarf shrub type through a short herb type to a tall herb type near Betsele, N. Sweden (data are fromGiesler et al., 1998^*,Nordin et al., 2001^–,Högberg et al., 2003^ǂ,2006^#, 2007a^§). Microbial biomass was estimated by the FE (fumigation extraction) method. Percentage organic N is water extractable organic N out of the sum of organic and inorganic N. Data are means (1.0 SE). Seasonal means (N = 3–4) except for soil pH and C/N (n = 9), total C and N (n = 24), and CECe

(n = 4–5).

Parameter Dwarf shrub

type

Short herb type

Tall herb type pH§(H2O)

4.0 (0.1) 4.6 (0.1) 5.3 (0.1)

C/N^§ 38.1 (2.4) 22.9 (1.1) 14.9 (0.3)

Total C (kg ha¹)^# 12,754 (435) 12,679 (510) 26,515 (1695)

Total N (kg ha¹)^# 312 (10) 529 (10) 1472 (94)

Organic N (%)^– 68 (6) 44 (7) 16 (1)

NH4+-N (salt extractable, kg ha¹)^# 0.294 (0.035) 0.604 (0.134) 1.962 (0.250)

NO3-N (salt extractable, kg ha¹)^# 0.132 (0.020) 0.274 (0.061) 1.122 (0.230)

CECe(mmolc kg ¹organic matter)* 316 (44) 657 (144) 831 (41)

Retention of¹⁵NH4+

(%)^# 79.0 (6.2) 35.0 (2.5) 23.0 (1.8)

Gross N mineralization rate (kg ha¹day ¹)^# 0.3 (0.1) 1.1 (0.3) 4.3 (1.1)

Net N mineralization rate (kg ha¹50 days¹) 0.0 (0.0) 0.1 (0.0) 0.6 (0.1)

Ratio fungi/bacteria^ǂ 0.44 (0.10) 0.18 (0.02) 0.02 (0.00)

Microbial C/N^ǂ 11.7 (2.0) 6.9 (1.6) 4.8 (1.3)

Microbial biomass C (kg ha ¹)^ǂ 284.3 (57.1) 223.3 (79.9) 494.3 (156.7)

Microbial biomass N (kg ha¹)^ǂ 24.3 (3.1) 31.7 (5.9) 102.3 (5.0)

Microbial C out of total soil C (%)^ǂ 1.42 (0.06) 1.45 (0.29) 1.45 (0.18)

Microbial N out of total soil N (%)^ǂ 7.6 (1.0) 6.9 (1.3) 7.3 (0.4)

Table 1

Pools and fluxes of N in typical boreal forests (Sponseller et al. (2016), except for denitrification flux in which case data are fromKlemedtsson et al.

(2005)).

Stocks of N kg ha ¹of N Internal turnover of N kg ha¹yr ¹of N

Trees 100–500 Plant uptake 15–50

Mor-layer soil 300–500 Litter-fall 5–25

Mineral soil 800–4000 Net mineralization 5–15

Total 1200–5000 Net nitrification 1–1

Inputs of N kg ha ¹yr¹of N Outputs of N kg ha¹yr ¹of N

N deposition 1–3 Leaching 0.5–1.5

Biological N2-fixation 1–3 Denitrification 0.1–1

(7)

Assuming 200 days of effective N2-fixation and N mineralization per year, the internal N cycling may reach 20–200 kg N ha ¹- year ¹, which is one-two orders of magnitude greater rates than all input fluxes of N combined. Similarly, such rates of N turnover, although they do not include the turnover of organic N compounds like amino acids, by far outcompete the N output fluxes (hydrolog- ical and gaseous pathways). Runoff of N is usually < 2 kg N ha ¹ year ¹(Kortelainen et al., 1997, 2006; Schelker et al., 2016) and consist to 70–90% of organic N species (Stepanauskas et al., 2000; Sponseller et al., 2014). As regards denitrification, Klemedtsson et al. (2005) estimated losses to be of the order 0.05–0.8 kg N ha ¹year ¹, i.e. comparatively small.

In conclusion, the internal turnover of N is much larger than the inputs and outputs. In fact, in most boreal forests the N cycle is relatively closed, and therefore constrained by rates and feedbacks within the local plant-soil ecosystem.

3.2. How important are plant litter characteristics?

Evidently plant species produce foliage and litter of highly vari- able food qualities for grazers and decomposing organisms; for which qualities differ between plant species and within species depending on site conditions (Berg and McClaugherty, 2008). The large site variations in soil N availability in boreal forests are associated with considerable variations in litter composition. But are these differences in litter quality indicators of variations in soil edaphic conditions rather than drivers of the site variability themselves?

Decomposition studies using litter bags show that N-rich litters decompose faster initially (typically for the first 0–2 years), but then slower than N-poor litters (Berg and McClaugherty, 2008). Lit- ter bag studies can cover up to around 5 years, after which the litter is so fragmented that loss through the mesh of the bags and ingrowth of mosses and lichens limit their use (Berg and McClaugherty, 2008). Such studies allow observations of processes in the litter physically above the mycorrhizosphere, but not in this zone where N availability to plants is expressed (see Section3.5).

Bomb¹⁴C dating of the horizons of the mor-layer inPinus sylvestris forest showed 5 ± 0.4 (mean ± 1.0 SE) years for the uppermost layer, the S-layer, 14 ± 1.0 years for the F-layer, where mycorrhizal tree roots and their fungi first appear, and 42 ± 0.5 years for the H- layer, the lowest horizon of the organic mor-layer (Franklin et al., 2003, see alsoLindahl et al., 2007and Section3.5.). So how relevant are observations of the initial stages of litter decomposition to our discussions about the nature of the N limitation in boreal forests?

If we accept the emerging view that virtually all soil organic matter can be degraded (Schmidt et al., 2011; Gleixner, 2013;

Lehmann and Kleber, 2015), it seems logical to assume that also most of the N may ultimately be released, and it matters less if the annual release can be attributed to a slow or a fast process from several or fewer cohorts (age-classes) of litter, respectively. Con- versely, if the litter quality is crucially important in this context, the system may not remain in steady-state provided the species composition and the quality of litter is changing.

We would like to stress the fact that in N-limited boreal forests, the microbial biomass in the F + H-layer contains around 6–9% of total N found in these layers and have a C/N ratio of around 8–12 as compared to around 34–41 for the OM (Table 2;Martikainen and Palojärvi, 1990; Giesler et al., 1998; Näsholm et al., 1998;

Högberg et al., 2003; Blaško et al., 2015; see also Bauhus and Khanna (1999)for data from other regions). Thus, the microbial biomass itself may be an important sink for N released in bioavail- able forms. In fact, the constancy of the ratio microbial N/total N across a wide range of soil C/N ratios (Table 2) suggests that it is the size and the turnover of the microbial N pool in the F + H hori-

zon that is the major bottleneck in terms of the N flow between soils and plants, rather than variations in initial C/N of the litter.

According toTable 2, the microbial N pool in the N-rich tall herb, TH, type is 4 times larger than in the N-poor dwarf shrub, DS, forest type, while the rate of gross N mineralization rate is 14 times higher, despite the nearly constant ratio of microbial N over total N. Note also that there is no net N mineralization in the DS type.

Hence, variations in litter C/N ratio (e.g., Berg and McClaugherty, 2008) or in potential influences of plant secondary compounds (e.g.,Smolander et al., 2012; Adamczyk et al., 2016), do not appear to greatly influence the relative size of the total soil N pool that is microbial N in the horizon where litter-derived N first becomes a substrate for mycorrhizal plant roots. This, and the facts that litter decomposition studies have not been able to trace the N in litter in detail from litter-fall to plant N uptake and that the effect of % N of litter on decomposition rate shifts over time (Berg and McClaugherty, 2008) have convinced us to leave out from this review the very rich literature on plant litter decomposition, and focus directly on processes in the vicinity of plant roots.

Before we do so, we must stress that the often strong correlations among the C/N ratios of fresh plant litter, presence of plant secondary compounds, and the availability of N do not necessarily imply that the N supply to plant roots is proximally controlled by these factors. An alternative view, which we will explore, is that soil N turnover in the root zone is physiological and controlled by the relations between the relative rates of C and N supplied in available forms to the soil organisms.

In systems rich in N, e.g., the TH forest vegetation type at Bet- sele (Table 2), the C/N ratio of microorganisms in the F + H layer is around 5, which indicates a total dominance of bacteria over fungi. There is also a much lower N sink strength (see data on immediate retention of¹⁵NH4+inTable 2) and a faster turnover of N than in the poorer forest types (see data on net and gross N mineralization inTable 2). Microorganisms in boreal forest soils, especially below the S-horizon, are likely C-limited (Ekblad and Nordgren, 2002; Demoling et al., 2008, see also Nazir et al.

(2010)for a general discussion on the rhizosphere) and this limitation should be strongest in the TH forest types as indicated by their microbial C/N ratios (Table 2). Under such conditions, the rate of decomposition of litter and associated release of N may well limit the rate of N cycling in the soil-plant system. In contrast, in systems with a higher ratio of fungi/bacteria, a much higher C/N ratio and slower turnover of microorganisms, the microbial N retention is much higher (Table 2).

Thus, we propose that the rate of litter decomposition may not always limit the rate of N-cycling and plant N uptake in boreal ecosystems. It likely does in the richer systems, where microbial immobilization in the root zone appears much less important (Table 2), but in the more N-limited systems recent findings suggest that the ECM fungi play a crucial role by sequestering much available N (Näsholm et al., 2013; Franklin et al., 2014;

Hasselquist et al., 2016), which they do not transfer to their tree hosts; this hypothesis will be elaborated more in detail in Section5.

3.3. How is N released in available forms and which forms of N are used by microbes and plants?

Ever since Sprengel formulated his mineral theory (Sprengel, 1826; von Liebig, 1840, 1855), plants have been considered the critical interphase between the inorganic and organic realms. This view prevailed, and still prevails in some quarters in spite of exper- imental evidence that plants may utilize organic N (Hutchinson and Miller, 1912; Brigham, 1917; Paungfoo-Lonhienne et al., 2012). Strong correlations between plant productivity and the occurrence of inorganic N, in particular NO3, as well as rapid Please cite this article in press as: Högberg, P., et al. Tamm Review: On the nature of the nitrogen limitation to plant growth in Fennoscandian boreal for-

(8)

growth responses of plants following supply of inorganic N formed a logical underpinning to this classic view. Occurrence of organic N in soil solution (e.g.Ivarson and Sowden, 1969) was not considered relevant for plant N nutrition, since it was assumed that microbes would always be superior competitors for N because their much larger surface to area ratio would effectively hinder any organic N from reaching plant roots (Schimel and Bennett, 2004;

Kuzyakov and Xu, 2013).

Recent studies have shown that N can be taken up by plants in a variety of forms, including peptides, amino acids, NH₄⁺and NO₃ (Näsholm et al., 2009). There is even evidence suggesting non- mycorrhizal plant roots may acquire N through uptake and diges- tion of microbes (Paungfoo-Lonhienne et al., 2010). This wide- ranging capacity of plants, with different types of mycorrhizas, suggest plant N nutrition to be constrained by access to N substrates, rather than by inherent plant competences to take up different forms of N.

The release of available forms of N can occur by a number of processes. Many organisms, notably microorganisms, exude exo- enzymes with the capacity to cleave polymers like proteins and peptides (Schimel and Bennett, 2004), but also plants exude pro- teases enabling for them to use complex N sources (Adamczyk et al., 2009). Release of N also occurs when microorganisms and plant roots die, which is followed by lysis, or when they experience a decreasing supply of C. Phage attacks have been known to drive turnover of microbial biomass in aquatic systems, but the ecological function of the highly abundant and diverse soil viruses is mainly unknown (Fierer et al., 2007; Srinivasiah et al., 2008). Co-evolution of soil bacteria and soil viruses suggests bacteria-phage interactions may also be pertinent to N turnover in soils (Gómez and Buckling, 2011). Grazing on bacteria by pro- tozoa (Clarholm, 1985), and on fungi by Collembola (Högberg et al., 2010) andAcari(Remén et al., 2008) accelerates the turnover of microbial N as it is followed by excretion of NH3by the soil animals.

The view that plant N nutrition also includes organic N sources rests on three fundamental observations: (i) all plants studied until now have been shown capable of absorbing and utilizing simple organic N forms such as amino acids, (ii) transporters mediating amino acid uptake have been identified both in mycorrhizal fungi and in plants, and expression of these transporters has been shown to respond to the presence of amino acids in the root media (see below), and (iii) dissolved organic N often constitutes a significant share of soil solution and runoff (Sponseller et al., 2014). Thus, on the one hand we have firm evidence indicating both the occurrence of the organic N substrates (e.g., amino acids) and the active and regulated uptake of them by plants from the soil. On the other hand, strong criticism against a significant role of organic N for plant N nutrition has been presented, arguing that plants are infe- rior competitors to soil microbes for organic N (Kuzyakov and Xu, 2013). The ECM fungi are particularly interesting in this context, given that they ensheath the vast majority of tree fine root tips, and although they are microorganisms they are conventionally ascribed the role of extensions of the tree root systems. We will elaborate on this notion below (3.5).

Studies using dual labelled organic N compounds were believed to offer a possibility to assess plant organic N uptake in the field (Schimel and Chapin, 1996; Näsholm et al., 1998, 2001). This approach was later criticized for being non-conclusive and for favouring plants rather than microbes by adding unrealistically high concentrations of labelled substrates (Jones et al., 2005;

Rasmussen et al., 2010). However, although the initial studies encompassing use of dual labelled tracers relied solely on bulk- analyses of stable isotopes in plants by Isotope Ratio Mass Spec- trometry, later studies that also used Gas Chromatography Mass Spectrometry provided proof for plant uptake of intact organic N

from soil (Näsholm et al., 2001; Öhlund and Näsholm, 2001;

Nordin et al., 2004).

Organic N nutrition of boreal forest plants is, based on the results discussed above, a reality. The quantitative importance of organic N for plants in this, or in any ecosystem is, however, still unknown. New insights into this field may come from studies of the genetic underpinnings of plant N uptake and assimilation (Näsholm et al., 2009).

3.4. Methodological limitations and developments

Transporters mediating root uptake of various forms of N, including both NH4+

, NO3, amino acids and peptides, are present in plants (for overview see Näsholm et al., 2009; Nacry et al., 2013). Not surprisingly, therefore, plants have been shown capable of absorbing and utilizing a wide range of N compounds. This potentially ubiquitous capacity of plants to acquire both organic and inorganic N sources suggest soil supply of N to be an important determinant of both the amounts and the chemical composition of plant N sources. To this end, recent developments of techniques to monitor soil N fluxes points to a much larger importance of organic N than previously recognized (Inselsbacher and Näsholm, 2012).

As discussed above, soils are dynamic systems, exhibiting rapid turnover rates of many compounds, not least N compounds. The same is true for the supply of plant photosynthate C, which fuels soil organisms with energy (Högberg and Read, 2006), which affects their N metabolism. Such rapid conversions are also charac- teristic, and maybe more intuitively recognized for endogenous compounds of an animal. The classical observer effect, i.e. the fact that any attempt to study a system also is at risk of affecting the system, was early documented within the field of neuroscience.

Hence, methods enabling low-invasive monitoring of neurotrans- mitters were developed, and a method later termed microdialysis was presented (Delgado et al., 1972). The method relies on a process termed induced diffusion and may best be explained as a form of reverse dialysis. The standard dialysis setup aims at removing substances from a solution while in the microdialysis procedure, the aim is to capture small molecules into a stream of pure water.

The small size of the probe (a standard setup uses a probe with a 0.5 mm10 mm (diameterlength) membrane), means disturbance of the study system can be kept to a minimum. Recovering N compounds through induced diffusion will also likely give more relevant information on soil N availability compared to measurements of soil solution concentrations. This is because flux rates rather than concentrations determine plant N capture. A full account of the use of microdialysis to study soil N dynamics is given byInselsbacher et al. (2011).

Several investigations using the microdialysis technique to study soil N fluxes in boreal forests have recently been published (Inselsbacher and Näsholm, 2012; Inselsbacher et al., 2014;

Oyewole et al., 2016). These studies collectively show that compared to soil extraction with either H2O or a salt solution, microdialysis sampling yields much lower fractions of inorganic N, in particular NH4+, and substantially larger fractions of amino acids.

Thus, while NH4+

strongly dominated in conventional extracts, in N poor soils accounting for 80 % of the pool of inorganic N and amino acids, amino acids dominated the diffusive fluxes captured with microdialysis, accounting for 80 % of the N flux (e.g.

Inselsbacher and Näsholm, 2012). A method comparison suggests this discrepancy results from breakdown of organic N and/or release of NH4+after soil sampling and during the subsequent sieving and extraction procedures (Inselsbacher, 2014).

Sampling, sieving and storage of soil samples also affects the supplies of C and N from organic substrates to soil microorganisms, and in particular the supply of photosynthate from plants to their mycorrhizal fungi and other organisms in the mycorrhizosphere Please cite this article in press as: Högberg, P., et al. Tamm Review: On the nature of the nitrogen limitation to plant growth in Fennoscandian boreal for-

(9)

(Högberg et al., 2003; Högberg, 2004; Högberg and Read, 2006).

Severing of fungal mycelium disrupts the cytoplasmic transport and may cause death and lysis of cells. The termination of the supply of C to ECM fungi weakens their strength as sinks for nutrients like N. Thus, there is a need to develop non-intrusive methods for studies of microbial physiologyin situ. This insight has stimulated us to make larger-scale stable isotope labelling directly in the field (e.g.,Högberg et al., 2010; Näsholm et al., 2013).

The modern approaches as well as the now more traditional all have their inherent weaknesses. Progress has undoubtedly been made, but one should always be cautious and regard any numbers derived by these methods as proxies of actual quantities or process rates. We find that the greatest values of these numbers are found when they are used for comparison, such as in studies of vastly different systems like the forest types at Betsele (Table 2).

3.5. Where in the soil profile is N supplied to trees and other plants?

We would like to repeat that the root zone and its extension, the mycorrhizosphere including the extramatrical mycorrhizal hyphae with its associated microbes, is the focal site of expressions of the balance between the supply of N and plant N uptake. In boreal forests mycorrhizal plant roots are not present in the uppermost part, the S-layer of the organic mor-layer (where most of the litter decomposition takes place), but become abundant first in the lower part of the mor, the F- and H-layers. Mor-layers are described as consisting of a superficial (S) layer of mosses, or lichens and plant litter, followed by the F-horizon (a Swedish word for decomposition was historically incorrectly translated to fer- mentation, hence F) in which plant remains can still be identified as such, and finally a ‘‘humus” (H) horizon in which the organic material is amorphous (Romell and Heiberg, 1931). Please observe that the internationally more common description of the uppermost layer as a litter layer (L-layer) is rarely appropriate in the open boreal forests on infertile soils, which commonly allow considerable penetration of light through tree canopies and thus a vital and continuous cover of mosses or lichens on the ground.

These grow through and become mixed with above-ground litter, which complicates studies of decomposition using the litter-bag method.

An observer gently removing the S-layer with its mosses or lichens and litter will observe the underlying F-horizon with an abundance of roots and dense fungal mycelial networks. Recent molecular studies have shown that saprotrophic fungi dominate in the litter in the S-horizon, which seems to lack ECM fungi (Lindahl et al., 2007; Clemmensen et al., 2013). The ECM fungi attain an almost total dominance in the F + H-horizons and further downwards (Lindahl et al., 2007; Clemmensen et al., 2013). We propose that this shift in dominance is associated with a diminishing availability of C in easily degradable forms to saprotrophic fungi as the litter they decompose is degraded and moved down the profile when new litter is added on the top, and as mosses and lichens grow upwards through the fragmented litter. Further- more, while saprotrophs experience a diminishing supply of easily degradable C-compounds, they become poorer competitors for N in relation to the ECM fungi, which receive sugars, high-quality C substrates, directly from their tree hosts. The appearance of prolific growth of fungal mycelia in the F-horizon occurs in the context of the shift in dominance from saprotrophs to ECM fungi observed in a study of decomposing needle litter (Lindahl et al., 2007).

Here, we would like to make an analogue with the competition for phosphorus between saprotrophic fungi and ECM fungi as studied in laboratory mesocosms byLindahl et al. (2001). In this, and similar studies, ECM fungi receiving photosynthate from pine seedlings were competing with saprotrophs supplied by C from wood blocks for P supplied as radioactive³²P to either the saprotroph

or the ECM fungus when they came in close contact. The subsequent movement of the tracer towards the mycorrhizal mycelium and its symbiotic plant or to the saprotroph was monitored by autoradiography. By manipulating the C supply rate to the saprotroph by varying the size of the wood block Lindahl et al. could demonstrate that the saprotroph became a stronger competitor for P when its C supply was larger.

Competition for N should follow the same principle, i.e. the sink with the greatest C supply should be the stronger sink. However, this competition will be more difficult to study, because radioactive N isotopes are extremely short-lived (most long-lived is¹³N with a half-life of c. 10 min). Such studies would need to be based on the stable isotope¹⁵N and destructive sampling.

Competition for N between microbes and plants is a classic theme in soil science with agronomic perspective, but also in soil ecology in general (e.g.,Kaye and Hart, 1997). The role of mycorrhizal fungi in this context is particularly difficult to establish.

Are they mere extensions of the plant root systems or is their origin as saprotrophic microbes still expressed to some extent? A recent analysis (Lindahl and Tunlid, 2015) concluded that while ECM fungi evolved from saprotrophs their saprotrophic capacities are in general not much expressed. Some ECM species can produce important exo-enzymes, and hence oxidize organic matter by so called ‘brown-rot’ Fenton chemistry or through the action of

‘white-rot’ peroxidases.Lindahl and Tunlid (2015)suggested that ECM fungi thus improved their access to N from organic matter, but that the associated release of C compounds would likely be less as important C sources for these fungi, but more important to saprotrophs.

The distribution of the stable isotopes of N (here expressed as per mil deviations,d¹⁵N, from natural abundance of¹⁵N in relation to that in the standard, atmospheric N2) in soil profiles provide interesting evidence about N processes and spatial separation of the processes without any disturbance of the studied system (Högberg et al., 1996; Högberg, 1997; Lindahl et al., 2007;

Hobbie and Högberg, 2012). Typically, in boreal forests the tree foliage is depleted in¹⁵N relative to the deeper soil horizons. This difference,

e

, is often up to around 10‰(e.g.,Högberg et al., 1996;

Sah et al., 2006; Blaško et al., 2015) and is created by N isotope fractionation as N is taken up by ECM fungi which become enriched in¹⁵N, but pass isotopically lighter N to their host plants (Högberg et al., 1999; Hobbie and Högberg, 2012). It should be stressed that the change in isotopic composition with increasing soil depth reflects a change in the isotopic mass balance of soil N. The difference

e

is influenced by the extent to which ECM fungi are active in tree N uptake; if the role of ECM is diminished, e.g., by decreasing the C supply to the roots and ECM fungi by high N additions (see further below),

e

decreases, but can increase again if the N load is terminated and the functional role of ECM fungi in N- limited forest is restored (Högberg et al., 2011).

Interestingly, when Lindahl et al. (2007) studied fungi on degrading pine litter, they observed no change in thed¹⁵N of the litter as long as it was colonized by saprotrophic fungi only, i.e.

during the first 3 years, which shows there was no change in the isotopic mass balance. A rapid increase in d¹⁵N started after 10 years when the ECM fungi had become dominant. The high enrichment in¹⁵N of deeper soil horizons thus suggests that N in ECM fungi may be an important precursor for more slowly turning over soil N pools (Högberg et al., 1996; Clemmensen et al., 2013).

In line with the revised view of so called recalcitrance (Schmidt et al., 2011), the observed¹⁵N enrichment might be reinforced by recycling of ECM N by ECM fungal mycelium and associated microorganisms (Gleixner, 2013). If saprotrophic fungi were the proximal cause of the change ind¹⁵N, by supplying¹⁵N-depleted N to the trees, and leaving¹⁵N-enriched N in the soil, the very high d¹⁵N of ECM fungi in relation to both plant and saprotroph N Please cite this article in press as: Högberg, P., et al. Tamm Review: On the nature of the nitrogen limitation to plant growth in Fennoscandian boreal for-